Automatically identifying domain variables based on data dependence graph

Experience in reengineering legacy system shows one shouldn´t assume the availability of the system documents; even if they are available, the documents may be inconsistent with the code. The software code becomes more reliable source than any documentation to get business rules. So, it is necessary to extract the business rules from source code. Domain variables can be mapped to important objects in the application domain. Identifying domain variables is a significant step in extracting the business rules from the source code. This paper proposes a solution to identify the domain variables automatically from the source code by applying data dependence graph (DDG). The solution contains three steps: generating the DDG of legacy system; identifying pure domain variables; identifying all domain variables which affect output domain variables; domain variables management. This solution been applied to a large complex financial legacy system which has proven to be successful.